M2 Acsyon (Local et MIX)

The course Online Optimization, Learning, and Games (O2LG) will be done on 10 weeks. There are 3 hours per week. 

The goal of online learning is the study of decision-making in changing environments - and, more specifically, to learn how to make more informed decisions that lead to better rewards over time. As a result, this field has found a wide range of applications, from traffic planning and online advertisement placement, to web ranking and the training of adversarial neural nets.

This course is intended to provide an introduction to online optimization and learning from a game-theoretic viewpoint. We will cover the basics of learning in finite games (focusing on the multiplicative/exponential weights algorithm and its properties), and the mainstay algorithms for online optimization in continuous games (online gradient descent, follow the regularizer leader / online mirror descent, etc.). We will also discuss the impact of the information available to the learners, as well as a range of concrete applications to machine learning and operations research.